The most important factors of successful trading strategy are the decisions to sell or buy. We propose multi-classifier system for decision making in algorithmic trading, whose training is carried out in three stages. At the first stage, features set is calculated based on historical data. These can be oscillators and moments that used in technical analysis, other characteristics of time series, market indexes, etc. At the second stage, base classifiers are trained using genetic algorithms, and optimal feature set for each of them is selected. At the third stage, a voting ensemble is designed, weights of base classifiers are selected also using genetic algorithms. However, the usage of genetic algorithms requires considerable time for computing, so the proposed system is implemented in a parallel environment. Testing on real data confirmed that the proposed approach allows to build a decision-making system, the results of which significantly exceed the trading strategies based on indicators of technical analysis and other techniques of machine learning.
When an asset-pricing model is claimed to explain a cross-section of portfolio returns, it should do so both within one asset class and across different asset classes. This paper illustrates that this is not always the case using the CAPM and Asness, Moskowitz and Pedersen (AMP, 2013) models applied to momentum and value portfolio returns as examples. Apparently, on one hand, the CAPM is almost as good as the AMP model in explaining the portfolio returns across asset classes, but on the other hand, the AMP model is almost as bad as the CAPM in explaining these returns within one asset class.
Our research is devoted to trade strategy’s profits and study of financial anomalies in stocks pricing. We analyze Momentum (and Reversal) strategies construction that is based on historical prices of assets. The main feature of the momentum strategy is that past stocks relative return (higher or lower than mean return or benchmark set) is used for selecting assets in portfolio.
The accent in our paper is made on revealing the nature of momentum and reversal (or contrarian) effects over time periods up to one year through the analysis of two basic determinants of abnormal profits of arbitrage portfolios of different design: cross-sectional variance of mean returns (rational explanation) and time-series predictability of asset returns (irrational explanation according EMH). The analyzed period embraces, from January 2006 to December 2014. Our research of Russian stock market has shown that, considering the choice of portfolio design (temporal windows for selecting stocks for portfolio and investment, and weight of stocks in the portfolio) and stock sample for constructing strategies (the sample should include major companies with liquid stocks) momentum and reversal effects do take place. Momentum profit is demonstrated in short-term strategies (3 to 6 months), while reversal effect is marked for ultra-short (less than a month) and long periods (11–12 months). Profit decomposition shows that the component responsible for rational explanations is statistically significant and its weight prevails in most momentum strategies with investment period not exceeding 9 months.
Carry trades consistently generate high excess returns with high Sharp ratios, but are subject to crash risk. I take a closer look at the link between the carry trade returns and the stock market to understand the risks involved and to determine when and why currency crashes happen. Every period, I sort currencies of developed and emerging economies by their interest rates and form portfolios to diversify the idiosyncratic risk. First, I find a strong negative relationship between portfolio returns and skewness of exchange rate changes. In fact, skewness and coskewness with the stock market have a much greater explanatory power in the cross-section of excess returns than consumption and stock market betas. But separating the market beta into upside and downside betas improves the validity of the CAPM significantly. Downside beta has a much greater explanatory power than upside beta, and it correlates with coskewness almost perfectly. This means that carry trades crash exactly in the worst states of the world, when the stock market goes down. After controlling for country risk, the downside beta premium in the currency market is comparable to that in the stock market and equals 2-4 percentage points p.a. I also find that country risk proxies well for the downside beta and skewness. This suggests that there is unwinding of carry trades and a “flight to quality” when the stock market plunges, and that lower interest rate currencies serve as a “safe haven”. Finally, I estimate even higher downside betas of the top portfolios and I find an even greater explanatory power of the downside beta in the early 2000s. The growing volume of carry activities might have contributed to the closer link between the currency and the stock markets.
The aim of this article is to prove the evidence of cross sectional momentum effect in Russian stock market within the variety of momentum strategy design elements and disclosure of the momentum effect nature.
Some currencies persistently move together with the stock market and crash in periods of market downturns or high volatility, while others serve as a “safe haven”. In this paper, I study whether or not countries’ macroeconomic characteristics are systematically related to the market risk of their currencies. I find that the market risk is not random, especially on the downside, and it can be predicted by macroeconomic variables. Moreover, the market risk has increased significantly since the 2000s, and its predictability also increased. The real interest rate has the highest explanatory power in accounting for the cross-section of currency market risk. Currencies of countries with high local real interest rates have high market betas, especially downside betas, while low real interest rate currencies are immune to stock market changes. Nominal interest rates also have some explanatory power, but only to the extent to which they correlate with the real interest rates. Other variables considered seem to be irrelevant.
High momentum returns cannot be explained by risk factors, but they are negatively skewed and subject to occasional severe crashes. I explore the timing of momentum crashes and show that momentum strategies tend to crash in 1-3 months after the local stock market plunge. Next, I propose a simple dynamic trading strategy which coincides with the standard momentum strategy in calm times, but switches to the opposite contrarian strategy in one month after a market crash and keeps the contrarian position for three months, after which it reverts back to the momentum position. The dynamic momentum strategy turns all major momentum crashes into gains and yields average return, which is about 1.5 times as high as the standard momentum return. The dynamic momentum returns are positively skewed and not exposed to risk factors, have high Sharpe ratio and alpha, persist in different time periods and geographical markets around the globe.